Mobile robots, and especially autonomous or semi-autonomous mobile robots, frequently have to cope with the problem of localizing themselves in a new or changing environment for which a predetermined map is not available. To overcome this difficulty, simultaneous localization and mapping (SLAM) algorithms have been developed and can be used to build a map of the surroundings of a robot in real time and to compute a localisation of the robot based on this map.
For instance, EP 2 058 720 describes a method for generating a three dimensional map using structured light emitted by a robot toward a ceiling.
However, this solution has a number of drawbacks.
In particular, the interaction of a semi-autonomous robot with a human operator often involves the use of a predetermined map of the environment, which the operator uses to guide the robot or to indicate a specific destination to the robot.
Unfortunately, this predetermined map often differs from the environment map generated by the robot in real-time. These differences can, for instance, arise from non-reference elements such as boxes, furniture, doors, or the like, which can be moved or can change their appearances over time and will thus prevent the robot from being able to associate the predetermined map with the environment map.
The instant invention has notably for object to improve this situation.